RealSign: Sign Language Recognition System

RealSign is a computer vision application designed to interpret sign language gestures into text. Utilizing the YOLOv11 architecture, the system provides real-time inference capabilities suitable for accessibility tools and communication interfaces.

System Overview

• Application Name: RealSign
• Model Architecture: YOLOv11 (Nano)
• Deployment Environment: Streamlit / Hugging Face Spaces
• Primary Function: Optical Character Recognition for Hand Gestures (A-Z)

Key Features

• Real-Time Inference: Low-latency processing of video frames for immediate feedback.
• Dual Input Modes: Supports both direct webcam feed and static image file uploads.
• Confidence Metrics: Displays probability scores for each detection to ensure reliability.
• Responsive Interface: Professional UI design adaptable to various display resolutions.

Technical Architecture

The system is built upon a modular Python stack:

1. Core Inference Engine: Ultralytics YOLOv11 trained on a dataset of 87,000+ annotated images.
2. Frontend Framework: Streamlit for web-based rendering and state management.
3. Image Processing: OpenCV and PIL for matrix manipulation and pre-processing.

Installation and Usage

Prerequisites

• Python 3.9 or higher
• pip package manager

Local Deployment

1. Clone the Repository

   git clone [https://github.com/YOUR_USERNAME/RealSign.git](https://github.com/YOUR_USERNAME/RealSign.git)
   cd RealSign

2. Install Dependencies

   pip install -r requirements.txt

3. Execute Application

   streamlit run app.py

Cloud Deployment

This application is configured for deployment on containerized environments such as Hugging Face Spaces.

• System Dependencies: Requires libgl1 (configured in packages.txt).
• Python Dependencies: Listed in requirements.txt.

License

This project is distributed under the MIT License.